THE JOURNAL OF BIOLOGICAL CHEMISTRV 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 264, No. 22, Issue of August 5, pp. 13109-13113,1989 Printed in U. S. A.

Carotenoid Biosynthesis in Photosynthetic Bacteria GENETIC CHARACTERIZATION OF THE RHODOBACTER CAPSULATUS CrtI PROTEIN*

(Received for publication, March 9, 1989)

Glenn E. Bartley and Pablo A. ScolnikS From the Du Pont Plant Science Group, Central Research and Development Department, Wilmington, Delaware 19880-0402

Carotenoids are photoprotective pigments present in many photosynthetic and nonphotosynthetic orga- nisms. The desaturation of phytoene into phytofluene is an early step in the biosynthetic pathway that in the photosynthetic bacterium Rhodobacter capsulatus is mediated by the product of the crtl gene. Here we report the sequence of this gene and the identification of CrtI as a membrane protein of approximate M, 60,000. Mutant strains with 5-fold lower or 10-fold higher levels of CrtI with respect to wild type have only small differences in their carotenoid content, in- dicating that the cellular concentration of CrtI is not a limiting factor in carotenoid biosynthesis. However, a correlation was found between the levels of CrtI and the formation of a photosynthetic antenna system.

Carotenoids are synthesized by bacteria, plants, fungi, and algae (1). Although their primary function is photoprotection (2), they are also precursors of vitamin A in humans (3) and of abscisic acid in plants (4). The first two steps in the biosynthetic pathway are mediated by soluble enzymes (5), whereas the remaining proteins of the pathway, which are currently the focus of our research, are membrane-bound (6). The dehydrogenation reaction that converts phytoene into phytofluene is common to all carotenogenic organisms, and it is the first biosynthetic step mediated by membrane proteins (6). Although these proteins can be solubilized with detergents and reconstituted in an active form (6), at the present time this approach only yields partially purified preparations. We have used previously a genetic approach to characterize the genes for carotenoid biosynthesis in the photosynthetic bac- terium Rhodobacter capsulatus, an organism for which effi- cient genetic tools have been developed (7). In that work we assigned biochemical functions to the R. capsulatus genes for carotenoid biosynthesis on the basis of the analysis of carot- enoids accumulated in uiuo and in vitro (8,9) and determined that crtl mediates the dehydrogenation of phytoene. In this paper we report the nucleotide sequence of the crtl gene and the characterization of the CrtI protein.

MATERIALS AND METHODS

Growth Conditions-R. capsulatus cells were grown under photo- synthetic conditions in ammonia-free minimal medium (RCVBNF) supplemented with threonine (10 mM) or ammonia as described (10).

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

to the GenBankTM/EMBL Data Bank with accession number(s) The nucleotide sequence(s) reported in this paper has been submitted

504969. $ To whom correspondence should be addressed Du Pont Experi-

mental Station, P.O. Box 80402, Wilmington, DE 19880-0402.

Matings were done as described (9). The wild-type strain was SB1003, the crtl point mutation used was BPY69, and the interposon insertion strains were described previously (8).

DNA Sequencing-Sequence was determined on both strands of pBLUESCRIPT (Promega, Madison, WI) derivatives by the chain termination method (11) using both commercially available and cus- tom-made primers. For sequence analysis we used the Wisconsin sequence analysis package (12).

Plasmid Construction-For expression of a CrtI-derived peptide in Escherichia coli we cloned the 1084-bp’ NdeI-BarnHI fragment from CrtI (Fig. 1) into a T7 translation vector (13). The resulting plasmid (pT701) was transformed into E. coli BL21 (DE3) and expression was induced as described (13). For expression in R. capsulutus a 2400-bp NruI-BglII fragment carrying crtl and a segment of crtB (8) was cloned between the NdeI (converted to a blunt end with T4 DNA polymerase) and the BamHI site of pNF3 (10).

Antibody Production and Immunoblots-Sections of NaDodS04- polyacrylamide gel electrophoresis gels containing the polypeptide expressed from pT701 were cut, frozen, and ground to a powder in liquid nitrogen. Antibodies were raised in New Zealand White rabbits by Hazelton Research Products, Denver, PA, and purified according to Olmstead (14). For immunoblots we used the ProtoBlot AP kit from Promega (Madison, WI) according to the manufacturer’s in- structions. The intensity of the reactive bands was visually compared with standards obtained with purified CrtI polypeptides expressed in E. coli.

Pigment Characterization-Bacteriochlorophyll was determined from acetone-methanol (7:2) extracts according to Cohen-Bazire et al. (15), and carotenoid according to Scolnik et al. (16).

RESULTS

Nucleotide Sequence of the crtl Gene-Based on insertional mutagenesis, nuclease S1 protection analysis, and comple- mentation of point mutations, we had determined previously (8) that crtl starts downstream of the NruI site (bp 6) and continues for about 1.4 kilobase pairs. The nucleotide se- quence of crtl and the deduced peptide sequence of the only open reading frame that conforms to a compiled codon usage for R. capsulatus (data not shown) is shown in Fig. 1. There are three in-frame ATG codons in the first 150 bp that could serve as start codons. Based on R. capsulatus codon usage and on homology to the equivalent gene from Neurospora crassa,2 we locate the putative translation start at the ATG at bp 88. This ATG codon is not preceded by a canonical Shine-Dal- garno (17), but the sequence 80 AAG could base pair with the 3’ end of the 16 S rRNA. The predicted M , of the protein is 57,997. Analysis of hydrophobicity (18) showed three regions with the potential for interacting with the membrane (Fig. 1). The terminator TGA overlaps with an ATG codon which is preceded by a Shine-Dalgarno sequence; this could correspond to the start of the crtB gene (8). The NruI site used to construct the interposon insertion SB203N (red phenotype)

The abbreviations used are: bp, base pair(s); LH-I and LH-11, light-harvesting antenna I and 11, respectively; NaDodS04, sodium dodecyl sulfate.

G . E. Bartley and P. A. Scolnik, manuscript in preparation.

13109

13110 R. capsulatus Crtl Protein

TGA I &Ill

// I

SB203N red

a I I SB203E SB218

blue-green blue-green

B 1 TTTTGTCGCG AGACAGCGTC GACAGTTGTA ATCGGAATTG ACACCTATCA TCCCCCCAAT GCAACCTGAA ACTACCGAAG -

8 1 AAACCATGTC CAAGAACACA GAAGGTATGG GTCGCGCCGT TGTCATCGGT GCCGGCCTTG GCGGTCTTGC TGCAGCGATG M S K N T E G M G R A V V I G A G L G G L A A A M 25

1 6 1 CGGCTGGGCG CAAAAGGTTA CAAGGTGACG GTCGTCGATC GTCTGGATCG GCCCGGCGGG CGTGGCTCTT CGATCACCAA R L G A K G Y K V T V V D R L D R P G G R G S S I T K 5 2

2 4 1 GGGCGGGCAT CGATTCGACC TTGGGCCGAC GATCGTGACG GTGCCCGACC GGCTGCGCGA GCTTTGGGCC GATTGCGGGC G G H R F D L G P T I V T V P D R L R E L W A D C G 7 8

3 2 1 GCGATTTCGA CAAGGACGTG AGCCTTGTGC CGATGGAGCC CTTCTACACC ATCGATTTCC CCGATGGCGA GAAATACACC -

R D F D K D V S L V P M E P F Y T I D F P D G E K Y T 1 0 5

4 0 1 GCTTACGGCG ATGACGCCAA GGTCAAGGCC GAGGTGGCGC GGATCAGCCC CGGCGATGTC GAGGGCTTCC GCCATTTCAT A Y G D D A K V K A E V A R I S P G D V E G F R H F M 1 3 2 -

4 8 1 GTGGGACGCC AAGGCCCGTT ATGAATTCGG CTATGAAAAC CTCGGCCGCA AGCCGATGAG CAAGCTGTGG GACCTGATCA W D A K A R Y E F G Y E N L G R K P M S K L W D L 1 158

5 6 1 AGGTTCTGCC GACTTTCGGC TGGCTGCGCG CCGACCGCTC GGTCTATGGC CATGCCAAGA AGATGGTGAA GGACGACCAC K V L P T F G W L R A D R S V Y G H A K K M V K D D H 1 8 5

6 4 1 CTGCGCTTCG CGCTGTCGTT CCATCCGCTT TTCATCGGCG GCGACCCCTT CCATGTGACG TCGATGTATA TCCTCGTCAG L R F A L S F H P L F I G G D P F H V T S M Y I L V S 2 1 2

7 2 1 CCAGCTCGAA AAGAAATTCG GCGTGCATTA CGCGATCGGC GGCGTGCAGG CGATTGCCGA TGCGATGGCC AAGGTGATCA Q L E K K F G V H Y A I G G V Q A I A D A M A K V I 2 3 8

8 0 1 CCGATCAGGG CGGCGAGATG CGCCTGAACA CCGAGGTCGA CGAGATCCTG GTCTCGCGTG ACGGCAAGGC CACGGGCATC -

T D Q G G E M R L N T E V D E I L V S R D G K A T G I 2 6 5

8 8 1 CGGCTGATGG ACGGCACCGA GCTTCCGGCG CAGGTTGTCG TCTCGAACGC CGATGCGGGC CACACCTACA AGCGTCTCTT R L M D G T E L P A Q V V V S N A D A G H T Y K R L L 2 9 2

9 6 1 GCGCAACCGC GACCGCTGGC GCTGGACCGA CGAGAAGCTC GACAAGAAGC GCTGGTCGAT GGGGCTTTTC GTCTGGTATT R N R D R W R W T D E K L D K K R W S M G L F V W Y 318

1 0 4 1 TCGGCACCAA GGGTACGGCC AAGATGTGGA AGGATGTGGG TCACCACACC GTCGTCGTCG GGCCGCGCTA CAAGGAACAT F G T K G T A K M W K D V G H H T V V V G P R Y K E H 3 4 5

1 1 2 1 GTGCAGGACA TCTTCATCAA GGGCGAGCTG GCCGAGGACA TGAGCCTTTA TGTCCACCGT CCCTCGGTCA CTGATCCGAC V Q D I F I K G E L A E D M S L Y V H R P S V T D P T 3 7 2

1 2 0 1 CGCGGCGCCG AAAGGCGACG ACACCTTCTA CGTGCTTTCG CCGGTGCCGA ACCTCGGCTT CGACAATGGC GTGGACTGGT A A P K G D D T F Y V L S P V P N L G F D N G V D W 3 9 8

1281 CGGTCGAGGC CGAGAAATAC AAGGCCAAGG TGCTGAAAGT GATCGAGGAA CGGCTGCTTC CGGGGGTTGC CGAAAAGATC S V E A E K Y K A K V L K V I E E R L L P G V A E K I 4 2 5

1 3 6 1 ACCGAGGAAG TGGTCTTCAC GCCGGAAACC TTCCGCGACC GTTATCTCTC GCCGCTGGGC GCGGGCTTCT CGCTGGAACC T E E V V F T P E T F R D R Y L S P L G A G F S L E P 4 5 2 -

1 4 4 1 GCGGATCCTG CAATCGGCCT GGTTCCGCCC GCATAACGCC TCGGAAGAGG TGGACGGGCT TTATCTGGTC GGCGCGGGCA R I L Q S A W F R P H N A S E E V D G L Y L V G A G 4 7 8

1 5 2 1 CCCATCCGGG CGCCGGTGTG CCCTCGGTGA TCGGCTCGGG CGAGCTTGTC GCGCAGATGA TCCCGGATGC GCCGAAGCCC T H P G A G V P S V I G S G E L V A Q M I P D A P K P 5 0 5

1 6 0 1 GAGACCCCCG CGGCGGCTGC GCCCAAGGCC CGGACGCCCC GGGCCAAGGC GGCGCAATGA E T P A A A A P K A R T P R A K A A Q ’ 5 2 5

FIG. 1. Genetic physical map, nucleotide sequence, and deduced polypeptide sequence of the crtl gene. A , physical map shows above the line the restriction sites at which interposon insertions were constructed and the location of initiation (ATG) and termination ( E A ) codons. The names of the interposon mutant strains and their color phenotype are shown below the line. The color phenotype of wild-type and crtZ mutants is, respectively, red and blue-green. B, nucleotide sequence of crtl numbered from bp 1 and deduced peptide sequence

bp 6), EcoRI (GAATTC, bp 505), Sal1 (GTCGAC, bp 838), and BamHI (GGATCC, hp 1445). The sequence in the one-letter code. Bars on top of the nucleotide sequence indicate relevant restriction sites: NruI (TCGCGA,

CCGATG (bp 352) was changed by site-specific mutagenesis to CATATG (NdeI). Stretches of peptides likely to form transmembrane domains (18) are underlined.

R. capsulutus CrtI Protein 13111

consensus

FIG. 2. Occurrence of His residues in hydrophobic areas of CrtI compared with the putative bacteriochlorophyll-binding antenna sequence. His residues are numbered on the right.

A pT701 T7P-

\ i Ndel Barn HI

1 2 c”-- U

cl cc- - - 4 1 m

60kDa

FIG. 3. Overexpression of CrtI in E. coli and R. capsulatus. A, plasmid pT701 containing a segment of crtl under control of the phage T7 promoter (T7P) and NaDodS0.-polyacrylamide gel electro- phoresis of extracts from E. coli cells carrying pT701 stained with Coomassie Blue. 10 pg each of soluble (lane 1 ) or insoluble (lane 2 ) fractions were run in a 10% gel. Location of the crtl-derived polypep- tide is indicated. E, plasmid pGBN3.2 containing crtl under control of the nij promoter (nijP). Black boxes and thin lines indicate, respectively, R. capsulatus and vector DNA. Below, immunoblot of R. capsulatus membrane fractions (10 pg/each) run in a 10% NaDodS0,- polyacrylamide gel electrophoresis gel, blotted onto nitrocellulose, and probed with the anti-CrtI antibody. Lane 1, SB1003; lane 2, SB203N; lanes 3 and 4, BPY69 (pGBN3.2). For lanes 1,2, and 4 the nijP was activated by growing cells under nitrogen fixing conditions. For lane 3, nijP was repressed by adding ammonia to the medium (10).

is approximately at 50 bp upstream of the start of transcrip- tion: whereas the EcoRI and Sal1 sites used, respectively, to construct SB203E and SB218 (blue-green phenotype) are in

Please note that the approximate start of transcription was er- roneously reported in the text of Ref. 8 as being at 200 bp downstream of the NruI site, whereas it is a t about 50 bp (see Fig. 3 of same reference).

the coding region. It has been postulated that the structural polypeptides of

the R. capsulatus antenna system interact with bacteriochlo- rophyll through His residues located in transmembrane a- helical domains (19,ZO). The common motif for these residues is Ala-X-X-X-His (19). Conserved His, but not Ala, residues were observed in antenna polypeptides of different photosyn- thetic bacteria (20, 22). There are four His residues in hydro- phobic areas of CrtI. Three of them contain either Ala or Gly, which is also a small side chain residue, at position -4 respect to His (Fig. 2). Using site-specific mutagenesis, Bylina et al. (21) have shown that the replacement of -4 Ala to Gly in a LH-I structural polypeptide does not alter the formation of this antenna complex.

Overexpression of a Segment of crtZ in E. coli and Zmmu- nological Detection of CrtZ in R. capsulatus-To obtain an anti-CrtI antibody we cloned a 1084-bp segment of crtZ into a T7 expression vector (Fig. 3). Induction of expression re- sulted in the synthesis of a 41-kDa insoluble protein (Fig. 3). An antibody generated against this protein recognizes a 60- kDa membrane protein in the wild-type strain SB1003 (Fig. 3). This is consistent with the information derived from the nucleotide sequence. The protein remained bound to mem- branes washed with 1 M NaC1, and the apparent size was not affected by different heat treatments of the sample before electrophoresis or by including 8 M urea in the gel (data not shown).

CrtI was found only in low levels in SB203N (Fig. 3), and, as expected, it is absent in the interposon mutants SB203E and SB218 (data not shown).

Complementation of crtZ- Mutants and Overexpression of CrtZ in R. capsulatus-To further characterize the role of the CrtI protein we constructed pGBN 3.2, a plasmid in which the crtZ gene is under the control of a promoter for nitrogen fixation (nif). We mobilized this plasmid by conjugation ir?to the R. capsulatus mutant BPY69, a blue-green strain which contains a crtZ69 mutation (9). The use of a point mutation for complementation analyses of crtZ is necessary because interposon insertions in this gene are polar on the adjacent crtE gene (8). BPY69 (pGBN3.2) merodiploids are red even when the nif promoter is not induced (data not shown) and accumulate little CrtI (Fig. 3), but induction of the nif pro- moter led to a 10-fold increase in CrtI respect to wild type (Fig. 3).

We analyzed the pigment composition of cells grown under nitrogen fixing conditions (Table I). Compared to wild type, the strain SB203N has about 54 and 60% of, respectively, carotenoids and bacteriochlorophyll. BPY69 (pGBN3.2) has about the same concentration of carotenoids but 70% higher bacteriochlorophyll than the wild type. The ratio of bacteri- ochlorophyll to carotenoids is similar in SB203N and the wild-type strain, but it is about 60% higher in BPY69 (pGBN3.2). We did not observe accumulation of phytoene in SB203N (data not shown).

Formation of Antenna Complexes-R. capsulatus has two

TABLE I Carotenoid and Bchl content of different R. capsulatus strains

Cells were grown in RCVBNF supplemented with threonine ac- cording to “Materials and Methods.” Carotenoids and Bchl concen- trations are expressed in micrograms/mg protein. Bchl, bacterio- chlorophyll.

Strain Carotenoids Bchl Bchl/Carotenoids SB1003 2.6 f 0.10 10.4 f 0.12 4.0 f 0.22 BPY69 (pGBN3.2) 2.7 0.02 17.7 * 0.21 6.6 * 0.12 SB203N 1.4 k 0.12 6.0 f 0.07 4.3 k 0.43

13112 R. capsulatus CrtI Protein

- 1 LH-I1 LH-I

750 850 Wavelength (nm)

FIG. 4. Near-infrared absorption spectra of membranes from SB203N, SB1003, and BPY69 (pGBN3.2). Absorption maxima of LH-I and LH-I1 are indicated on top. Samples were normalized for the bacteriochlorophyll content.

photosynthetic antenna systems, LH-I and LH-11, both formed by structural proteins, bacteriochlorophyll, and carot- enoids (20). Under the low-light phototrophic growth condi- tions used in this work, LH-I1 is dominant. The infrared absorption maxima are 875 nm for LH-I and 800 and 850 nm for LH-I1 (20). SB203E and SB218 do not form LH-I1 (8), a result consistent with previously published work showing that mutations in the early steps of carotenoid biosynthesis affect the formation of this antenna complex (23). We have now examined the spectrum of SB203N and found that little LH- I1 forms in this strain (Fig. 4). In contrast, BPY69 (pGBN3.2) forms more LH-I1 than the wild type. We used electron microscopy to examine cells of all three strains and found no significant differences in cell size or degree of development of the photosynthetic membranes (data not shown).

DISCUSSION

This investigation shows that the R. capsulatus CrtI protein is an intrinsic membrane protein of approximate M, 60,000. The association with the membrane might be mediated by at least one of the three hydrophobic segments identified. Sev- eral of our results show that there is little correlation between the levels of CrtI and the concentrrkion of carotenoids. First, SB203N contains approximately 5-fold less CrtI than SB1003 (Fig. 3), but accumulates about 54% of the carotenoids of the wild type. Second, when the plasmid pGBN3.2 was introduced into cells of the crt1 mutant strain BPY69, the phenotype changed from blue-green to wild-type red, although the nif promoter was not induced, and no CrtI protein was detected in immunoblots (Fig. 3). Third, overproduction of CrtI with the nif expression vector (Fig. 2) did not result in an increase in carotenoid content with respect to wild type (Table I). However, there is a significant increase in the bacteriochlo- rophyll/carotenoid ratio (Table I).

Ultimately, characterization of the function of the individ- ual proteins of the carotenoid biosynthetic pathway depends

on our ability to reconstitute purified components into either natural or artificial membranes containing carotenoid precur- sors. At the present time, purification of these proteins in an active form cannot be easily achieved. The anti-CrtI antibody described here could be used to purify CrtI from R. capsulatus cells without having to rely on biosynthetically active systems. When the nucleotide sequence of other genes for carotenoid biosynthesis becomes available (28), the protocols described here can be used to characterize the remaining proteins of the pathway.

We have also investigated the formation of the photosyn- thetic antenna LH-I1 system in relation to the stoichiometry of CrtI. In photosynthetic bacteria there is a correlation between the presence of colored carotenoids and the formation of antenna systems (8, 23). The mechanism for this adaptive response, which allows the bacteria to limit the size of the photosynthetic antenna when photoprotective carotenoid pig- ments are absent, is not known. Carotenoids per se do not mediate the interaction, because an R. capsulatus mutant that lacks colored carotenoids can form low levels of LH-I1 (27). Our results with SB203N, which contains a significant amount of carotenoids but only low levels of LH-11, tend to support this lack of direct connection between the presence of colored carotenoids and the formation of LH-11. However, our observations on the formation of LH-I1 in the strains SB203N and BPY69 (pGBN3.2) (Fig. 3) suggest that there is a correlation between the levels of CrtI and the formation of LH-I1 complexes. In the case of overproduction of CrtI, the increased formation of LH-I1 is accompanied by an increase in the ratio of Bchl to carotenoids (Table I). There are two possible explanations for these phenomena. First, it is possible that changes in the stoichiometry of CrtI give rise to unspe- cific interactions, for instance changes in the properties of the photosynthetic membrane, that alter pigment concentra- tion and antenna formation. Second, it is possible that CrtI plays an accessory role in bacteriochlorophyll binding (24, 25, 26) and LH-I1 formation. The data currently available does not allow us to distinguish between these two possibilities.

Acknowledgments-We are grateful to D. Pollock (Cold Spring Harbor) for his early contributions, to N. Iusem (Du Pont) for the affinity purification of the anti-CrtI antibody, and to G. Armstrong (Berkeley) for communicating sequence information before publish- ing.

REFERENCES 1. Weedon, B. C. L. (1971) in Carotenoids (Mer O., ed) pp. 29-59,

2. Krinsky, N. I. in Carotenoids (Isler, O., ed) pp. 669-716, Birkhau-

3. Pitt, G. A. J. in Carotenoids (Isler, O., ed) pp. 717-742, Birkhauser

4. Zeevaart, J. A. D., and Creelman, R. A. (1988) Annu. Reu. Plant

5. Dogbo, O., Laferriere, A., D’Harlingue, A., and Camara, B. (1988)

6. Bramley, P. M., and Sandmann, G. (1987) Phytochemistry 26,

7. Scolnik, P. A., and Marrs, B. L. (1987) Annu. Reu. Microbial. 4 1 ,

8. Giuliano, G., Pollock, D., Stapp, H., and Scolnik, P. A. (1988)

9. Giuliano, G., Pollock, D., and Scolnik, P. A. (1986) J. Biol. Chem.

10. Pollock, D., Bauer, C. E., and Scolnik, P. A. (1988) Gene (Amst.)

11. Kraft, R., Tardiff, J., Krauter, K. S., and Leinwand, L. A. (1988)

12. Devereux, J., Haeberli, P. and Smithies, 0. (1984) Nucleic Acids

Birkhauser Verlag, Basel, Switzerland

ser Verlag, Basel, Switzerland

Verlag, Basel, Switzerland

Physiol. Plant Mol. Biol. 39 , 439-473

Proc. Natl. Acad. Sci. U. S. A. 8 5 , 7054-7058

1935-1939

703-726

Mol. & Gen. Genet. 213,78-83

261,12925-12929

65,269-275

Biotechniques 6 , 544-547

Res. 12,387-395

R. capsulatus Crtl Protein 13113

13. Studier, F. W., and Moffatt, B. A. (1986) J. Mol. Biol. 189 , 113- 21. Bylina, E. J., Robles, S., and Youvan, D. C. (1988) Zsr. J. Chem.

14. Olmsted, J. B. (1986) Methods Enzymol. 134, 467-472 22. Zuber, H. (1985) Photochem. Photobiol. 42, 821-844 15. Cohen-Bazire, G., Sistrom, W. R., and Stanier, R. Y. (1957) J . 23. Zsebo, K- M., and J. E. 37* 937-947

130 28, l -6

Cell. Comp. Physiol. 49, 25-68

Chem. 255,2427-2432

24. Tronrud, D. E., Schmid, M. F., and Matthews, B. W. (1986) J. 16. Scolnik, P. A., Walker, M. A., and Marrs, B. L. (1980) J. Bid. 25. Michel, H., Epp, o., and Deisenhofer, J. (1986) EMBo J , 5,

17. Shine, J., and Dalgarno, L. (1974) Proc. Natl. Acad. sei. u. s. A. 26, yeates, T. o,, ~ ~ ~ i ~ ~ , H., chirino, A., R ~ ~ ~ , D. c., ~ l l ~ ~ , J. p.,

18. Kyte, J., and Doolittle, R. F. (1982) J . Mol. Biol. 157, 105-132 and Feher, G. (1988) Proc. Natl. Acud. Sci. U. S. A. 85, 7993- 7997

19. Youvan, D. c., and Ismail, s. (1985) PrOC. Natl. SCi u. s. 27. Dorge, B., Klug, G., and Drews, G. (1987) Biochim. Biophys. Acta

20. Zuber, H., and Theiler, R. (1984) Hoppe-Seyler's 2. Physiol. 28. Armstrong, G. A., Alberti, M., Leach, F., and Hearst, J. E. (1989)

Mol. Biol. 188, 443-454

2445-2451

7 1,1342-1346

A. 82,58-62 892,68-74

Chem. 365,721-729 Mol. & Gen. Genet., in press